Firm demography and aggregate productivity growth. Lars Fredrik Andersson

Transcription

1 Firm demography and aggregate productivy growth Lars Fredrik Andersson

2 Abstract This paper examines the dynamic relation between firm demography and aggregate productivy firm in Sweden during the period By using enterprise demography data, the interaction between micro dynamics and aggregate productivy growth is discerned. The result of the decomposion shows that the contribution of entry and ex of firms on aggregate productivy growth is small but still posive for the total economy, suggesting that entering firms are more productive than exing firms. The paper shows that the productivy growth whin the continuing firms is the key factor of the aggregated productivy growth. ITPS, Swedish Instute For Growth Policy Studies Studentplan 3, SE Östersund, Sweden Telephone: +46 (0) Fax: +46 (0) ISSN For further information, please contact Lars Fredrik Andersson Telephone +46 (0)

3 1 Introduction The present study examines the dynamic contemporary relation between firm demography and aggregate productivy growth in Sweden using panel data from 1997 to This is believed to be a particularly apt topic of research given that the creation of new business and decline of unproductive firms often is regarded as key characteristic of business dynamics. Indeed, economic theory predicts that successful firms will grow, while less successful firms will shrink or die and policy makers often believe that intense firm creation is important for productivy growth and job creation. 1 The analysis of firm demography, especially births and deaths of enterprises, could contribute to the understanding of the growth process in several key regards. First, the aggregate productivy growth is affected by the large ongoing reallocation of outputs and inputs across enterprises. Secondly, the intensy to wch this reallocation varies between sectors could have an impact on the industry productivy growth. Third, most of the reallocation could reflect firm dynamics rather than industry dynamics. 2 In accordance, structural change and aggregate productivy growth decomposed by the shift-share method, concordant report small shift s. 3 Fourth, there are large differences in productivy levels and growth across firms in the same industry. 4 Given that firm demography has a significant impact on aggregate productivy growth, this may skew the analysis of the relative importance of different factors of production. In particular, the of market dynamics may be erroneously attributed to technological progress when in fact the is a question of increased economic efficiency, because of increased usage of better technology. By offering a large sample of firms and, not only manufacturing industry, the contribution of firm demography on aggregate productivy growth can be discerned. Focusing on the key aspects of firm demography, this paper addresses the issue of the relative importance of entry and ex of firms, and whin-firm growth and market dynamics. Using a comprehensive dataset of Swedish firms, wh firm-level data on employment and value added, we try to establish the relative importance of resource reallocation and firm dynamics on productivy growth. Since tentative results shows that surviving new firms have lower productivy but a higher productivy growth that existing firms, the following hypothesis are addressed: (1) New firms have a limed impact on productivy growth, because of a high infant mortaly rate. Dynamics of existing firms is more important for productivy growth than firm entry. (2) Reallocation of resources to the most productive firms is an important contributor to technological advancement. The rest of the paper is structured as follows. Section 2 describes methodology and data applied in the study. Section 3 outlines the contribution of entry, ex and incumbent on aggregate productivy growth. Section 4 concludes. 1 OECD (2002); Ahn (2001);Bertelsman and Doom; Caves (1998);Geroski (1995), 2 Foster, Haltiwanger and Krixan (1998) 3 Peneder (2003); Fagerberg (2000); Timmer and Szirmai (2000). 4 Geoski (1995). 3

4 2 Method and data In discerning productivy, the most frequently applied measures are labour productivy and multi-factor productivy. As the latter is accounting for the distinct s of capal/labour inputs together wh technological progress, is often seen as favourable. 5 However, due to the lack of proper capal stocks on micro level, has not been possible to measure total factor productivy. Therefore, the second best measure i.e. labour productivy is employed in the study. In this regard, should also be noted that labour productivy has advantages: the risk of error measurement is reduced and the correlation wh other productivy measures is high. 6 The lerature on the linkages between firm dynamics and productivy growth offers a wide range of methods. One of the most frequently applied methods is decomposion, where the contribution of productivy components or factors is discerned. The ways to accounts for the factors behind productivy differs between the methods of decomposion chosen. Following the suggestions of Ahn (2001), the aggregate productivy in a given industry can be measured by a weighted average of each firm s productivy in the sector: LP t = i θ LP Where θ is the market share of the h firm, ln LP is the firm labour productivy. The employment or value added share may be used as weights. This study employes labour shares. In line wh the preferred method by Haltiwanger (1997) and Foster, Haltiwanger and Krizan, the aggregate LP growth can be decomposed as follows: 1. Whin Between Cross Entry Ex LP = θ LP + θ ( LP LP ) + θ LP + θ ( LP LP ) θ ( LP LP t iεc k iεc k t k iεc LP t k Where denotes to change between the k-year interval between the first year (t-k) and the last year (t). C, N and X refers to continuing, entering and exing firms. LP t-k is the aggregate productivy level of the industry. The contribution of the factors decomposed is interpretated as follows. (1) The firm productivy growth weighted by inial market shares is defined as whin. (2) The second term represent the between-firm component that reflects changing shares, adjusted for the average productivy. (3) This term represent a cross (covariance), that is posive when the market share growth for firms wh growing labour productivy. (4) The entry denotes the sum of differences between each firm s productivy and inial aggregate productivy, weighted by s market share. (5) The sum of differences between every existing firm s labour productivy and inial aggregate labour productivy, weighted by s market share represent the ex. In sum, the components defined in equation 3, involve deviations of firm-level labour iεn t k iεx k k t k 2. 5 Foster, Haltiwanger and Krizan (1998) 6 Hakkala (2004). 4

5 productivy from the inial productivy level of the industry. Therefore, the risk of over aggregating the of the between, entry and ex is reduced. Alternative methods for decomposing productivy growth, such as suggested by Campbell (1992) and Griliches and Regev (1992) is more sensive for the over aggregating problem of between, entry and ex terms. Moreover, the difficulty of interpretation the components distinctively is another reason for preferring the method defined in equation 3. 7 In this study, the contribution of continuers, entering and exing firm s is decomposed. By referring to firms according to these categories, is necessy to have robust definions. In this study, the definion employed by statistics Sweden is suggested to be appropriate to meet these needs. 8 The information on firm dynamics and statistics are supplied by the IFDB database, created by the Swedish instute for Growth Policy Analysis (ITPS). Wh this database is possible to use organisation numbers to identify firms and the firms have the same number of the period of study. An exing firm is defined as a firm wh that belongs to the industry in 1997, but where the organizational number is missing for the industry group in In accordance, an entering firm is defined wh an organizational number that belongs to the examined industry group in 2003, but not in Wh this database, is possible to discern the links between different entry and ex categories. Genuine entering and exing firm is separated from spl and mergers. Besides the use of organisational numbers as a key to identify firm dynamics, this study employs a demographic method denoted FAD (Dynamics of firms and working places). Wh the FAD method, comprehensive details on entering and exing firms can be uncovered. The risk of double counting a firm that change organisational number is avoided. The FAD method classifies firms as an entry, ex and incumbent firm depending on changes of labour force whin the firm. This firm dynamic data can be joined wh data on value added and number of employed (and also financial statements) by matching organisational numbers. In the present study, the decomposion of productivy will be conducted using FAD-data, in addion to using organisation numbers. 9 The dataset employed in the present study include information on value added and persons employed by the different categories of firms. The period of study is limed to , since the sample of firms included in the database changed in 1996/97. Subsequently, calculations based on the 1996 data can not be compared wh that form 1997 onwards whout invoking measurement problems. One problem of the dataset is to measure labour productivy from micro firms. Micro firms wh (<1 employed) and unincorporated enterprises (where the owner not is counted as employed) is left out to avoid these measurement problems. The risk is however, that the dynamic of micro firms is neglected. Still, wh zero employed is impossible to calculate labour productivy. In line wh the latter are firms wh negative value added excluded. To avoid mixing up different categories of entering and exing firms, only genuine entering and exing firms is included. Spls and mergers, wch falls between incumbents and entering/exing, are excluded. Based on these arguments is the sample delimed according to the following creria: (1) Firms wh less than one employed: (2) Firms wh negative value added: (3) Unincorporated enterprises. (4) Firms that merge and spl. All firms that meet these creria are excluded. 7 Heden (2005); Foster, Haltiwanger and Krizan (1998); Ahn (2001); Campbell (1992). 8 Andersson and Arvidson (2006) 9 Details of the FAD method is presented in appendix 2. 5

6 3 Results of productivy decomposion This study covers a sample of firms that are entering, exing and continuing over the present study period. The study covers firms from all industries. In table 1 below the number of firms by industry and by category is presented. Table 1. Number of entering, exing and continuing firms in sample. ISIC rev. 3 Sector Entering Exing Continuing Primary sector Manufacturing Construction and infrastructure Trade, hotel and restaurants Transport and communications Financial and business services Education, social services Total Source: IFDB database, own calculations The number of entering, exing and continuing firms over the period 1997 to 2003 is , and respectively. At a closer look we find that the entering firms are concentrated to sector 4 (trade, hotel and restaurants) together wh sector 6 (financial and business services). The table shows a similar distribution among the exing firms. About two thirds of the firms are located sectors 4 and 6. In difference, about one half of the continuing firms are located in sector 4 and 6. The continuing firms have a stronger presence in sector 2 (manufacturing) and sector 3 (construction and infrastructure). About 30 per cent of the continuing firms are located in the two latter sectors. In comparison, some 20 per cent of the entering and exing origin from sector 2 and 3. The differences in firm distribution are an important feature to note. Given that the productivy level and growth differ between sectors, the contribution of firm, origin from different sectors, also is likely to differ. To measure the contribution of firm dynamics on aggregated productivy growth, the decomposion of labour productivy is reported in the following section. Focus will be put into examining the seven sectors presented above. In addion, theses sectors will be disaggregated into industries to show how the contribution may differ at a more detailed level. And to put these results into a context, stylized facts on productivy growth and firm dynamics is outlined as background. Over the last fifteen years, the number of entering and exing firms has developed rapidly. Focusing on the previous, we find that the number of entering of firms almost has doubled. Back in 1985, some 23 thousand firms were created and in 2004, almost 42 thousand new firms emerged. Over this period, the entry rate has fluctuated considerably. In figure 1 below, the development of labour productivy and entry is outlined. The figure shows that the number of entering firms developed rapidly up until the economic crises of the 1990s. After the crises the firm creation gathered strength up until the late 1999s. In the end of the period, the growth of new firms has been more slowly, except for the final year. 6

7 Figure 1. Entry of firms and labour productivy, index 1995=100, Entry of firms Labour productivy Source: ITPS, Nyföretagandet i Sverige 2004 och 2005; SCB, Nyföretagandet i Sverige 1994/95; GGDC, total economy database. The figure shows that the growth of labour productivy and entry of firms over the period 1985 to 2004 equals to 2,1 and 3,0 per cent respectively. During the study period ( ), the growth of labour productivy is somewhat lower and the growth in entry is much lower during the study period compared to the previous period. Taking the period 1997 to 2003 into closer examination, we find that the most intense firm creation is taking place in financial and business services. An equal important sector is gross, retail trade, hotel and restaurants. Both sectors accounts for about 60 per cent of the population Surprisingly few firms are created in the manufacturing sector. This pattern resembles of that seen in table 1. Figure 2. Number of entering firms (all categories) by sector Education, social services Financial and business services Transport and communications Trade, hotel and restaurants Construction Manufacturing Source: IFDB, own calculations. Note: The primary sector (agriculture, fishing, forestry) is excluded. All categories of entering firms (genuine new, spls and mergers) are included. The development of exing firms by industry over the period 1998 to 2003 share much of the same features as that of entering firms. Most of exing firms is located in the financial 7

8 and business services sector together wh the sector of trade, hotel and restaurants. We also find that the overall number of exing firms have changed somewhat. In 1998, some 39 thousand firms left the market, while about 42 thousand abandon in The largest two sectors (trade and business services) accounted for 60 per cent of the population. Figure 3. Number of exing firms (all categories) by sector Source: IFDB, own calculations. Education, social services Financial and business services Transport and communications Trade, hotel and restaurants Construction Manufacturing Note: The primary sector (agriculture, fishing, forestry) is excluded. All categories of exing firms (genuine new, spls and mergers) are included. The number of entering and exing firms includes all genuine new firms together wh spls and mergers of firms. The latter two account for a proportion of 20 per cent of the entering firms. The development of spls and mergers follows much of the fluctuations seen in overall entry rate, although the amplude is lower. The overview shows that the number of entering and exing firms is changing slowly, compared to historical record. In comparison, we find that the growth in number of entry and ex was significant higher over the period 1985 to 1996, the gorwth rate was significant higher. The overview also shows that most of the firms (60 per cent) origin from two sectors (Trade, hotel and restaurants together wh financial and business services sectors. To account for how this business dynamics have affected the aggregated productivy growth, the following sector offers a decomposion of labour productivy growth in Sweden for the period 1997 to Before presenting the decomposion on the firm level a decomposion at the industry level first is conducted. The method employed is the tradional shift-share analysis. 10 The data cover industries at the 2 and 3-dig industry level. Labour shares have been used as weights. Labour productivy has been deflated by 2 and 3 dig industry price indices. 11 Figure 4 shows the results of the decomposion of labour productivy growth in Sweden in 1997 to The decomposion is presented for total economy, manufacturing and services sectors. We find that the total labour productivy was close to 20 per cent over the 10 The shift share method is presented in appendix Due to the lack of price indices, services are covered at the 2-dig level and manufacturing at the 3-dig level. Therefore is the shift-share based on 3-dig data on manufacturing data and 2-dig data on services. The data is derived from the IFDB database. 8

9 period (3,2 per cent per year) The results of the shift-share shows that the whin is of crucial importance for the aggregated productivy growth. The shift-s are small and negative. The impact of reallocation of labour between the sectors is small on total productivy growth. Figure 4. Shift-share analysis of labour productivy growth in Sweden (all industries) 50,0% 40,0% All industries 30,0% 20,0% Manfacturing 10,0% 0,0% Services -10,0% Labour productivy Whin Static Dynamic Source: IFDB, own calculations. The figure shows that the static is negative, suggesting that the labour force moves from industries wh high productivy level towards industries wh lower productivy level. Similarly the dynamic indicate that the labour force moves from industries wh high productivy growth to industries wh lower productivy growth. The service sector has the relatively largest negative shift s. In the manufacturing sector, the shift s may be overlooked. The manufacturing sector contributes most to the aggregated productivy growth. Following the result on the reallocation between industries, the following section decomposes the reallocation between firms. The following decomposion is based on micro data and the method is denoted by equation 2. In table 2 below, the result by sectors is shown. We find that the aggregated productivy growth over the period reached 22 per cent (3,3 per cent per year). We also see that the figure of aggregated labour productivy in table 2 is somewhat higher compared to that in figure 4 above, (22 per cent compared to 20 per cent). This is because of differences in the dataset, where mergers and spls are excluded in the firm data. Considering table 2, the result shows that the contribution on total productivy growth is concentrated to the manufacturing sector. The latter have a share of 66 per cent of the total contribution. Following the contribution list in table 2, we find that transport and communication is the second most important sector, having a share of 22 per cent. The third most contributing sector is trade, hotel and restaurants, representing a share of 20 per cent. Taken together, sector 2, 4 and 5 leave a contribution of more than 100 per cent implying that the other four sectors have a negative contribution. 9

10 Table 2. Labour Productivy decomposion by industries, labour shares as weights. ISIC rev. 3 Sector Labour productivy growth Whin Between Cross Net entry Contribution Primary sector 9,0% 0,1% 0,0% 0,0% 0,1% 0,2% Manufacturing 41,3% 14,8% -0,1% -0,4% 0,4% 14,7% Construction and infrastructure Trade, hotel and restaurants Transport and communications Financial and business services Education, social services -5,5% -0,7% 0,0% 0,0% 0,1% -0,5% 22,2% 3,6% 0,0% 0,0% 0,8% 4,4% 56,7% 4,6% 0,0% -0,3% 0,6% 4,9% -9,1% -1,3% 0,4% -0,1% -0,2% -1,3% -0,2% 0,1% 0,0% 0,0% -0,2% -0,2% Total 22,1% 21,2% 0,2% -0,9% 1,6% 22,1% Source: IFDB, own calculations. Studying the decomposed components closer, we find that the sectors wh a posive continuing (sum of whin, between and cross ) also have a posive net entry (entry ex ). The largest net entry comes from sector 4 and 5. The posive counterbalances the negative s from sector 6 and 7. The total of entry and ex is posive, implying that new firms have a higher productivy level than the ones closed down. The posive relation between net entry and continuing s are also indicated at the 2-dig industry level. In figure 5, a scatter plot of this relation is shown. The figure indicates that posive continuing s are correlated wh posive net entry s. Figure 5. Correlation between net entry/ ex and continuing s. 6,0% 5,0% 4,0% 3,0% 2,0% 1,0% 0,0% -1,0% -2,0% -3,0% -0,6% -0,4% -0,2% 0,0% 0,2% 0,4% 0,6% 0,8% 1,0% Source: IFDB, own calculations. A regression analysis shows that the impact is significant, when net entry is dependent and continuing is independent. One hypothesis could be that entering firms, by using new production technology, developing production methods, can utilize the whin 10

11 that is pertinent to incumbents. An analysis of the relation shows that the continuing has a significant impact on the net entry. Table 3. The relation between Result of regression analysis Variable Parameter estimates t Value Pr > t Intercept 0,0001 0,46 0,6503 Whin 0,0534 2,43 0,0206 Between 0,0723 0,59 0,5609 Cross 0,0329 0,27 0,7864 The regression results in table 3 should be seen as an indication of dependence between whin and net entry. Still, the result support the notion that in industries experiencing rapid technological change, the entry of new firms is important in fostering aggregated productivy growth. 12 However, to support the results, a more theoretically funded argumentation together wh more comprehensive empirical analysis is needed. Considering the development of the manufacturing sector, we previously noted that has the most significant impact on the total productivy growth. To account for this strong performance, the development of the manufacturing sector is examined more closely. Table 4. Labour productivy decomposion, manufacturing industry, labour shares as weights. ISIC rev. 3 Industry Labour productivy growth Whin Between Cross Net entry Contribution Food, drink & tobacco 14,2% 1,0% 0,0% -0,1% 0,1% 1,0% 2.2 Textiles, clothing and leather industry 4,6% 0,1% 0,1% 0,0% 0,1% 0,2% Wood products 23,1% 1,0% 0,0% 0,0% 0,1% 1,1% Pulp & paper products 25,3% 2,5% -0,4% -0,4% 0,0% 1,6% Printing & publishing -2,2% -0,1% 0,1% 0,0% 0,1% 0,1% Mineral oil refining -43,1% -0,4% 0,0% 0,1% 0,0% -0,4% Chemicals 126,5% 9,6% 0,7% 2,4% 0,0% 12,7% Rubber & plastics 9,9% 0,3% 0,0% 0,0% 0,0% 0,3% 2.9 Non-metallic mineral 26 products 15,3% 0,3% 0,0% 0,0% 0,0% 0,4% Basic metals 38,3% 1,8% 0,0% 0,3% 0,0% 2,1% Fabricated metal products -1,9% -0,2% 0,0% 0,0% 0,1% 0,0% Mechanical engineering 23,7% 3,0% 0,0% 0,1% 0,1% 3,2% Office machinery 2,5% 0,0% 0,0% 0,0% 0,0% 0,0% Electrical machinery 32,7% 0,8% 0,0% 0,1% 0,0% 0,9% Industry for electronics and communication equipment 33,3% 1,9% 1,2% 0,9% 0,2% 4,2% Precision instruments 2,4% 0,1% -0,1% 0,0% 0,1% 0,1% Industry for motor vehicles 97,2% 12,0% 0,1% 1,0% 0,1% 13,1% Other transport equipment 5,2% 0,1% 0,0% 0,0% 0,0% 0,1% Other manufacturing 12,2% 0,3% 0,0% 0,0% 0,1% 0,4% Manufacturing 41,3% 34,2% 1,7% 4,3% 1,2% 41,3% Source: IFDB, own calculations. 12 OECD

12 In table 4, the labour productivy growth is decomposed in relation to the aggregated productivy in the manufacturing industry, i.e. the contribution sum up to 41.3 per cent. Besides the development in the manufacturing sector, the transport and communication sector had a strong impact on the aggregated productivy growth for total economy. The transport and communication sector accounted for 22 per cent of the total productivy growth. In table 6 below is the sector contribution distributed on the 2-digt industry level. As shown, the communications sector has a strong impact on the sector figure. Indeed, almost half of the sector origin from the industry for communications. At a closer look we find that the whin is dominating, while the net entry is nonexistent. On the contrary, we find a significant net entry in inland transport together wh supporting and auxiliary transport activies. The result of the sector is scattered when comes to net entry s. It is difficult to discern a distinct pattern. Table 5. Labour productivy decomposion, transport and communication sector. ISIC rev. 3 Industry Labour productivy growth Whin Between Cross Net entry Contribution Inland transport 26,2% 7,5% -1,5% 1,2% 2,4% 9,6% Water transport 113,2% 6,6% 0,0% -0,4% 0,6% 6,8% Air transport 10,0% 0,1% 0,1% 0,1% 0,2% 0,4% Supporting and auxiliary transport activies 49,7% 11,5% 0,0% 0,6% 3,7% 15,8% Communications 108,9% 35,7% -2,6% -9,0% 0,0% 24,1% 5. Transport and communications 56,7% 61,5% -4,0% -7,6% 6,8% 56,7% Source: IFDB, own calculations. The development of labour productivy has not, however, been posive across all sectors in the economy. To find out more about the negative contribution recognised in table 2, the financial and business service sector, having the largest negatively contribution (-1,3% of total 22,1), is put under closer examination. The result is reported in table 6. Table 6. Labour productivy decomposion, financial and business services. ISIC rev. 3 Industry Labour productivy growth Whin Between Cross Net entry Contribution Financial instutions and insurance companies -2,3% 0,0% 0,0% 0,0% 0,0% 0,0% 70 Real estate activies 14,0% 4,7% -6,4% -1,9% -0,7% -4,2% 71 Renting of machinery and equipment 47,5% 1,0% 0,0% -0,1% 0,8% 1,7% 72 Computer and related activies -16,6% -2,1% 0,1% -0,4% 0,2% -2,2% Instute for research and development -5,9% -2,0% -0,2% 0,0% -2,3% -4,4% Financial and business services -9,1% 1,7% -6,5% -2,5% -1,9% -9,1% Source: IFDB, own calculations. In overall we find a relatively large impact of entry and ex in the sector. This follows from the result reported in table 1, showing a large number of entry and ex, compared to incumbents, i.e. high turnover rates. In detail we find that especially real estate activies 12

13 together wh instute for research and development have the most negative on the aggregated productivy growth of the sector. Obviously, instutes for research and development could have posively externalies that are not taking into account wh the method employed. Another explanation could be that measuring of labour productivy is mixed up due to the wide range of activies included (legal and technical and advertising together wh other business activies, nec.) in instute for research and development. This mix of activies makes correct measurements difficult, especially since detailed price indices are not accessible. Cautionary interpretations should also be applied for financial instutions and insurance companies as only a small number of companies are covered. The development of these covered companies should also be taken cautionary due to the difficult measurement of value added in banking and insurance. The activies associated wh financial and business services could also have posive s that are not measured by the method employed in this study. Considering finally the creation and destruction of firms in the financial and business services, we find that industries for renting of machinery and equipments and computer and related activies have posive net s. In the latter sector, the net entry is posive while the whin is negative. This result suggests that entering firms are more productivy than incumbents and exing firms respectively. In overall the productivy growth of the sector is more dependent of entry and exing, compared to the other sectors. 13

14 4 Concluding remarks This paper has examined the dynamic relation between firm demography and aggregate productivy firm in Sweden during the period By using enterprise demography data, the interaction between micro dynamics and aggregate productivy growth has been discerned. The result of the decomposion shows that the contribution of entry and ex of firms on aggregate productivy growth is small for the total economy. Indeed, the contribution of net entry is 1,6 per cent of total 22,1 per cent. Still, this result suggests that entering firms are more productivy than exing firms. The of net entry seem to play only a modest role in manufacturing sector, while boost the development of labour productivy in services sectors. The net is especially significant in domestic trade, hotel and restaurants together wh transport and communications. For the total economy is the net entry more important than shifts in reallocation of labour between incumbents. The results of the present study both show similaries but also differences wh reference to the body of lerature on firm dynamics and aggregate productivy growth. Importantly, this study support the result that a large part of the aggregate labour productivy is driven by what happens in each individual firm, while shifts in reallocation of labour form incumbents in decline to those that are growing play less role. 13 A number of studies have reported that the growth of labour productivy is boosted by the destruction of low productivy firms and the creation of high productivy firms. This study supports this notion in general. Although the contribution is small in the Swedish manufacturing section, the entry of new uns is important in fostering the growth of labour productivy in services sectors. The analysis of the relation between whin (used as a proxy for technological progress) and net entry also support the result that the entry of new firms contributes to the aggregated productivy growth. 14 In the present study is the contribution of net entry on aggregated productivy smaller than that reported in a number of other studies. This may be explained by a number of factors. A longer time period of study generally leads to greater emphasis on net entry. Moreover is the whin is relatively smaller when the decomposion is applied on multifactor productivy. The decomposion on labour productivy generally give the whin a greater role. This could be the of incumbents raising labour productivy by increasing capal intensy and or shedding of labour. The use of different decomposion method and obviously data sources could also have an impact on the result. Comparisons between different studies should be made carefully accordingly. 13 Ahn (2001); OECD (2001); Foster, Haltiwanger and Krizan (1998); 14 Ibid. 14

15 5 Reference Ahn (2001) Firm dynamics and productivy growth: A review of micro evidence from OECD counstris, OECD Economics department Working Paper 297, OECD publications, Paris. Andersson and Arvidson (2006) Företagens och arbetställenas dynamik (FAD) [The dynamic of firms and plants], Statistic Sweden. Baily, M. Hulten C. and Campbell, D. (1992) Productivy dynamics of manufacturing plants, Brooking papers on economic activy, Microeconomics, pp Bartelsman, E. J. G. and Doms (2000) Understanding productivy: lessons from longudinal micro databases, Journal of Economic lerature, Vol 38, September Caves, R. E. (1998) Industrial organization and new findings on the turnover and mobily of firms, Journal of economic lerature, Vol 36:4, pp Fagerberg J. (2000) Technological progress, structural change and productivy growth: a comparative study, Structural change and economic dynamics Vol. 11 pp Foster, Haltiwanger and Krixan (1998) Aggregate productivy growth: lessons from microeconomic evidence, NEBR working paper no Geoski P. A. (1995) What do we know about entry? International Journal of Industrial Organization, Vol. 13. pp GGDC, total economy database, [ ] Hakkala K. (2004) Corporate restructuring and labour productivy growth, Working paper no. 619, The research instute of industrial economics, Stocholm. Heden Y. (2005) Productivy, upskilling, entry and ex: evidence from the UK and the Swedish micro data, Thesis submted for the degree of PhD, Queen Mary universy of London. IFDB database (Database of individuals and enterprises in Sweden ), ITPS. ITPS, Nyföretagandet i Sverige 2004 och 2005 [Newly-started enterprises in Sweden 2004 and 2005], ITPS, Östersund. OECD (2002) OECD Small and medium enterprises outlook, OECD publications, Paris. OECD (2001) Productivy and firm dynamics: evidence from micro data in OECD economic outlook, chapter 7, OECD publishing, Paris. Peneder M. (2003) Industrial structure and aggregate growth, Structural change and economic dynamics vol. 14 pp SCB, Nyföretagandet i Sverige 1994/95 [Newly-started enterprises in Sweden 1994/95], SCB, Stockholm. Timmer M.P and Szirmai A. (2000) Productivy growth in Asian manufacturing: the structural; bonus hypothesis examined, Structural change and Economic dynamics vol. 11 pp

16 6 Appendix 1: the shift-share method The shift-share method provides a tool to account for how the aggregate productivy growth is mechanically linked to different growth of labour productivy and the reallocation of labour between industries. In accordance, the method relates directly to the two contrasting mechanism previous denoted structural burden or bonus to the aggregate productivy growth. Following the methodology suggested by Fagerberg (2000) or Pender (2003), we decompose the aggregate productivy growth into three separate s: LPT, t1 LPT Growth( LPT ) = LP (1)Whin shift (2) Static shift effekt (3) Dynamic shift n ( LPi,t1 LPi,t 1 ) i= 1 S i,t 1 T, t 1 + n i= 1, t 1 LP = i,t 1 (S LP i,t1 T,t 1 S i,t 1 ) + LP Where LP is labour productivy; t-1, base year; t final year; T, over industries I; S i share of industry I in total employment. (1) The first component, the whin, capture the change in labour productivy, under the constrain that not structural shifts have taken place and that each industries has maintained the share amount of shares in total employment as in the base year. (2) The second component, the static shift, is calculated as the sum of relative changes of labour across industries between t-1 and t, weighted by the inial value of labour productivy in t-1 (3) The third component, dynamic shift, is calculated as the sum of changes in labour shares and labour productivy of industries. n i= 1 (LP i,t1 i,t 1 )(S i,t1 S i,t 1 ) 16